docker-compose.yml

# Perfis:
#   local  — lake em ./data/lake (sem GCP)
#   cloud  — lake no GCS + BigQuery (requer credentials/ e .env)
#
# API:
#   docker compose --profile local up api-local -d
#   docker compose --profile cloud up api-cloud -d
#
# Pipeline (one-shot):
#   docker compose --profile local run --rm pipeline-local daily-sync
#   docker compose --profile cloud run --rm pipeline-cloud ingest-sofascore --history --all-teams
#
# Atalho: ./scripts/docker-dev.sh cloud up

x-app-image: &app-image
  build: .

x-app-runtime: &app-runtime
  <<: *app-image
  working_dir: /app
  environment: &container-paths
    LAKE_ROOT: /data/lake
    SOFASCORE_STATS_DIR: /data/lake/sofascore
    SOFASCORE_FEPT_DIR: /data/lake/fept
    WC_ARTIFACT_DIR: /data/lake/artifacts/wc_predictor
    PORT: "8080"

services:
  api-local:
    <<: *app-runtime
    profiles: [local]
    ports:
      - "${API_PORT:-8000}:8080"
    environment:
      <<: *container-paths
      LAKE_PRIMARY: local
      LAKE_SYNC_BQ_ON_WRITE: "false"
    volumes:
      - ./data/lake:/data/lake
    restart: unless-stopped

  api-cloud:
    <<: *app-runtime
    profiles: [cloud]
    ports:
      - "${API_PORT:-8000}:8080"
    env_file:
      - path: .env
        required: false
    environment:
      <<: *container-paths
      LAKE_PRIMARY: cloud
      LAKE_SYNC_BQ_ON_WRITE: "${LAKE_SYNC_BQ_ON_WRITE:-true}"
      GCP_PROJECT: "${GCP_PROJECT:-}"
      BQ_DATASET: "${BQ_DATASET:-sports_news_lake}"
      GCS_BUCKET: "${GCS_BUCKET:-}"
      GOOGLE_APPLICATION_CREDENTIALS: /app/credentials/${GCP_CREDENTIALS_FILE:-beanalytic-dev.json}
    volumes:
      - ./credentials:/app/credentials:ro
      - ./data/lake/artifacts:/data/lake/artifacts
      - ./data/lake/fept:/data/lake/fept
    restart: unless-stopped

  pipeline-local:
    <<: *app-runtime
    profiles: [local]
    entrypoint: []
    command: ["daily-sync"]
    environment:
      <<: *container-paths
      LAKE_PRIMARY: local
      LAKE_SYNC_BQ_ON_WRITE: "false"
    volumes:
      - ./data/lake:/data/lake

  pipeline-cloud:
    <<: *app-runtime
    profiles: [cloud]
    entrypoint: []
    command: ["daily-sync"]
    env_file:
      - path: .env
        required: false
    environment:
      <<: *container-paths
      LAKE_PRIMARY: cloud
      LAKE_SYNC_BQ_ON_WRITE: "${LAKE_SYNC_BQ_ON_WRITE:-true}"
      GCP_PROJECT: "${GCP_PROJECT:-}"
      BQ_DATASET: "${BQ_DATASET:-sports_news_lake}"
      GCS_BUCKET: "${GCS_BUCKET:-}"
      GOOGLE_APPLICATION_CREDENTIALS: /app/credentials/${GCP_CREDENTIALS_FILE:-beanalytic-dev.json}
    volumes:
      - ./credentials:/app/credentials:ro
      - ./data/lake/artifacts:/data/lake/artifacts
      - ./data/lake/fept:/data/lake/fept